Kinship Verification through a Forest Neural Network

2025-04-26

AI Summary

To address the limitations of single-image modeling and strong reliance on joint image training in kinship verification, this paper proposes Forest-GNN, a graph neural network based on a forest-structured topology that explicitly models cross-generational facial feature correlations between parents and children—without requiring end-to-end joint image input. A novel progressive center loss fusion mechanism is introduced to jointly optimize multiple loss objectives, enabling discriminative metric learning in the feature space. By transcending the representational constraints of single-image encoders, Forest-GNN achieves state-of-the-art performance on KinFaceW-II (improving average accuracy by 1.6%) and near-optimal results on KinFaceW-I. The source code is publicly available.

Technology Category

Computer Vision: Generative Adversarial Networks (GANs) for VisionMachine Learning: Graph-based Machine LearningConstraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphsSemantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMsSearch and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Early methods used face representations in kinship verification, which are less accurate than joint representations of parents' and children's facial images learned from scratch. We propose an approach featuring graph neural network concepts to utilize face representations and have comparable results to joint representation algorithms. Moreover, we designed the structure of the classification module and introduced a new combination of losses to engage the center loss gradually in training our network. Additionally, we conducted experiments on KinFaceW-I and II, demonstrating the effectiveness of our approach. We achieved the best result on KinFaceW-II, an average improvement of nearly 1.6 for all kinship types, and we were near the best on KinFaceW-I. The code is available at https://github.com/ali-nazari/Kinship-Verification

Problem

• Improving kinship verification accuracy using joint facial representations
• Applying graph neural networks for enhanced face representation analysis
• Introducing new loss combinations to optimize network training performance

Innovation

• Uses graph neural network for face representations
• Introduces new combination of losses gradually
• Achieves best results on KinFaceW-II dataset